IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM...

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 13, NO. 5, SEPTEMBER/OCTOBER 2007 1497

Optical-CDMA in InPRonald G. Broeke, Jin Cao, Chen Ji, Sang-Woo Seo, Member, IEEE, Yixue Du, Student Member, IEEE,

Nick K. Fontaine, Jong-Hwa Baek, John Yan, Francisco M. Soares, Fredrik Olsson,S. Lourdudoss, Senior Member, IEEE, Anh-Vu H. Pham, Senior Member, IEEE, Michael Shearn,

Axel Scherer, and S. J. Ben Yoo, Senior Member, IEEE

Abstract—This paper describes the InP platforms for photonicintegration and the development on these platforms of an opticalcode division multiple access (O-CDMA) system for local area net-works. We demonstrate three building blocks of this system: anoptical pulse source, an encoder/decoder pair, and a threshold de-tector. The optical pulse source consists of an integrated collidingpulse-mode laser with nearly transform-limited 10 Gb/s pulses andoptical injection locking to an external clock for synchronization.The encoder/decoder pair is based on arrayed waveguide gratings.Bit-error-rate measurements involving six users at 10 Gb/s showederror-free transmission, while O-CDMA codes were calibrated us-ing frequency resolved optical gating. For threshold detection af-ter the decoder, we compared two Mach–Zehnder interferometer(MZI)-based optical thresholding schemes and present results ona new type of electroabsorber-based MZI.

Index Terms—Arrayed-waveguide gratings (AWG), cross-correlation frequency-resolved optical gating (XFROG), InP plat-form, Mach–Zehnder interferometer (MZI), optical code divisionmultiple access (O-CDMA), optical thresholding, photonic inte-grated circuit.

I. INTRODUCTION

PHOTONIC integration has been maturing to a level wheresystem scale functionality can be integrated on a single

substrate. Most suitable for advanced photonic integration is theInGaAsP/InP material system, because of its tunable bandgapacross the 1.3- and 1.5-µm wavelength communication windowsof optical fiber and, therefore, the ability to integrate active andpassive waveguides. An InP platform supporting active-passiveintegration provides three key functionalities on a single sub-strate: passive optical waveguiding (with bend radii down to50–500 µm for compact design), phase shifting, and amplifi-cation/detection of light. Platforms usually add one or morefunctionalities like electroabsorption modulators, polarizationconverters, spot-size converters, on-chip mirrors, or/and electri-cal high-frequency (HF) phase and HF amplitude modulators.Several InP platforms, using different techniques, have beenreported: quantum well (QW) intermixing [1] enables active-passive integration without regrowth, but often at the expense

Manuscript received December 18, 2006; revised June 15, 2007. This workwas supported in part by DARPA and SPAWAR under Agreement N66001-01-1and under Contract HR0011-04-1-0054.

R. G. Broeke, J. Cao, C. Ji, S. W. Seo, Y. Du, N. K. Fontaine, J. H. Baek, J.Yan, F. Soares, A.-V. H. Pham, and S. J. B. Yoo are with the Electrical Engi-neering Department, University of California, Davis, CA 95616 USA (e-mail:[email protected]).

F. Olsson and S. Lourdudoss are with the School of Information and Com-munication Technology, Royal Institute of Technology, S-16440 Stockholm,Sweden.

M. Shearn and A. Scherer are with the Division of Engineering and AppliedScience, California Institute of Technology, Pasadena, CA 91125 USA.

Digital Object Identifier 10.1109/JSTQE.2007.902854

of increased waveguide loss due to band-tail edge absorptionand the introduction of lattice defects by the implanted ions.Butt-joint selective-area (SA) regrowth in combination with adouble-etch technique [2], [3] provides flexibility in the posi-tioning and type of the active layer and allows for compact de-sign by high-contrast waveguide bends. However, it is difficultto control the regrowth of large-step butt-joints and to accuratelycontrol the etch depth required for the low-contrast waveguides.Integration by symmetric twin-waveguide structure [4] does notrequire a regrowth, but introduces possible optical absorption inthe taper region. Also, the effective index of both waveguidesshould match accurately to limit the taper lengths.

In our approach [5], we use a hydride vapour phase epitaxy(HVPE) lateral regrowth for simultaneous waveguide passiva-tion and planarization. It allows for a noncritical waveguideetching similar to buried waveguide processes. In this paper,we report on two platforms with respect to the developmentof an integrated O-CDMA system: one platform for low fre-quency (LF) components and a new platform for high frequency(HF). The additional tradeoffs in chip design introduced by HFcomponents for rapid code reconfiguration justifies a separateHF platform. Furthermore, we reduced waveguide loss by im-proved HVPE lateral growth conditions for more uniform side-wall coverage. We employed frequency-resolved optical gatingfor more accurate optical-phase calibration of the O-CDMAcoding/decoding and generating shorter pulses from the mode-locked laser. Finally, we introduced a new Mach–Zehnder inter-ferometer (MZI) thresholder design.

We develop the O-CDMA system using a parallel-track ap-proach. On the first track, we realize a table-sized optical free-space/fiber-based system to demonstrate feasibility, principle ofoperation, and to lay the groundwork for system scale charac-terization. For example, the free-space SLM-based 32-user ×10 Gb/s O-CDMA system [6]. On the other track, we developthe O-CDMA system on an InP platform. The parallel track ap-proach provides an environment, in which we progress towardfull integration by inserting InP devices incrementally into aproven operating system architecture. Full integration in semi-conductor is a prerequisite for robust and large-scale practicaldeployment.

II. INTEGRATION TECHNOLOGY

In this section, we give a brief overview of the integratedwaveguides in our two platforms: for LF and HF (10–40 GHz).

First, we consider the LF O-CDMA platform. Fig. 1 shows afully integrated O-CMDA system consisting of a transmitter andreceiver section. The transmitter consists of a mode-locked laser(MLL) integrated with a spectral-phase encoder. The receiver

1077-260X/$25.00 © 2007 IEEE

1498 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 13, NO. 5, SEPTEMBER/OCTOBER 2007

Fig. 1. Schematic picture of a fully integrated O-CDMA system.

Fig. 2. Cross-sectional view of waveguides. (a) Passive waveguide. (b) Phaseshifter. (c) SOA. (d) Laser with FIB facets. (e) EA.

consists of a decoder integrated with an MZI-based thresholddetector (MZI-TD).

Fig. 2 shows the cross-sectional view of the waveguide typesin the O-CDMA devices. The orientation of the cross sections issuch that the waveguide widths are displayed in the x-direction(lateral), thickness in the y-direction, and the propagation direc-tion of the guided light in the z-direction (longitudinal). Fig. 2(a)displays a passive waveguide of a 500-nm-thick quarternaryQ(1.15) waveguiding (film) layer sandwiched in InP. Lateralregrown Fe-InP layers passivate the sidewalls of the waveguideand provide electrical isolation. In addition, the lateral regrowthplanarizes the waveguide for subsequent metallization steps.The typical waveguide width is 3 µm, which is a compromisebetween low propagation loss (2.5 dB/cm) and single-modewaveguiding (avoiding the symmetrical second-order mode).Fig. 2(b) shows a phase shifter, whose refractive index can bechanged by injecting carriers in the guiding layer (forward bias),or by applying an electrical field across the Q layer (reversebias). The metal (AuGeNi/Au) and highly doped InGaAs layeron top provide a low-resistance n-contact for electrical connec-tion to an outside current or voltage source, for instance, by wire

Fig. 3. Traveling-wave phase modulator. (a) Lateral cross section. (b) Longi-tudinal cross section.

bonding in packaged devices. Semiconductor optical amplifiers(SOA) [Fig. 2(c)] have multiquantum well (MQW) layer on topof the film layer (MQW1). The MQW1 consists of six wellsof 7-nm-thick Q(1.25) InGaAsP and 6-nm-thick InGaAs barri-ers. The designed gain peak lies at 1550 nm. Fig. 2(d) showsa longitudinal cross section of an SOA with etched facets. Thefocused ion beam (FIB) etched facets provide optical feedbackfor the SOA, and create a laser suitable for integration of thelaser with other devices on a single substrate A second type ofMQW layer, MQW2 [Fig. 2(e)] provides a peak gain at around1490 nm and has been optimized for electroabsorption (EA)modulator operation; for instance, for data modulating a pulsetrain from a 10-GHz mode-locked laser, or for phase shifting inan MZI structure, as described later in this paper.

Fast code reconfiguration in the encoder/decoder demandsHF traveling-wave phase modulators. These modulators need tobe optimized for low propagation loss of HF electrical fields,velocity matching of the optical group and electrical phase ve-locity, and maximum overlap between the optical fields and theelectrical field.

Fig. 3 shows the cross sections of a traveling wave modula-tor. The most pronounced change with respect to the stack fromFig. 1(a) is the replacement of the n-doped InP substrate witha semi-insulating (SI) substrate. The n-contacts are placed onthe top side of the chip, on top of an n-InP buffer layer. Thisconfiguration reduces electrical propagation loss by reducing theamount of electrical field in n-doped layers. A benzocyclobutene(BCB) layer on both sides of the Fe-InP provides additional sup-port for the metal contact layer on top. The SiO2 layers providean electrically isolated cohesion layer for the metal. Removalof the n-buffer layer between adjacent modulators isolates theirn-contacts.

BROEKE et al.: OPTICAL-CDMA IN INP 1499

Fig. 4. Cross sections. (a) Phase shifter. (b) SOA. Cross sections of high-frequency EA modulator. (c) Lateral. (d) Longitudinal.

The epitaxy on an SI substrate consists of a 500-nm-thickn-InP buffer layer, a 500-nm Q (1.15) layer, a 2.0-µm p-InPlayer with gradually increasing p-doping toward the top, and a100-nm-thick InGaAs contact layer. A RIE etch through the Qlayer defines the waveguides. A lateral Fe-InP regrowth pas-sivates the waveguides and planarizes the surface. In order toreduce the overlap of the electrical field with the laterally re-grown Fe-InP, a wet chemical etch reduces the Fe-InP widthto 4 µm on both sides, which is wide enough not to affect theoptical guiding properties of the waveguide. Furthermore, sincethe microwave electrical field tends to have a lower propagationvelocity compared to the optical field, increasing the electricalfield confinement in air (by thinner Fe-InP) improves velocitymatching. The etching stops at the top of the n-buffer layer toenable the n-contacts.

Fig. 3(b) shows the same modulator, but in a longitudinal crosssection through the waveguide center. The segmentation of thep-contact geometry provides velocity matching of the electricalphase with the optical group velocity. The segmentation consistsof ion-implantation to locally destroy the conductance of the p-layer and a BCB layer to avoid metal-semiconductor contact onthe ion-implanted segments.

Due to the SI substrate, all electrical contacts need to beplaced at the top side of the device. Fig. 4(a) shows an LF phaseshifter in the HF-platform and Fig. 4(b) shows an SOA. In caseof current injection into the chip, the resistance from the p- ton-contact should be minimized. The resistance of the n-bufferto the n-contact can be limited to a few ohms by placing then-contact as close as possible to the waveguide and by makingit sufficiently doped and thick. Fig. 4(c) and (d) displays anEA with high-speed electrical connections. It is similar to thetraveling-wave modulator, except that the EA is shorter (upto 300-µm long), which makes velocity matching much lesscritical. Hence, no ion implantation is required.

III. O-CDMA ON INP

A SPECTS O-CDMA system employs spectral encoding ofultrashort optical pulses from a mode-locked laser for data trans-mission. Spectral coding in the transmitter spreads pulses intime at the expense of peak power. The decoder only recon-structs the ultrashort high-peak-power pulse if the encoder code

Fig. 5. SEM image of a 10-GHz CPM laser with integrated active-passivewaveguides.

and decoder code are conjugates of each other. (The decoderand encoder are similar devices.) The receiver, after the de-coder, discriminates correctly decoded pulses from incorrectlydecoded pulses by nonlinear threshold detection. The receiverdetects a “1” bit only if a pulse has sufficient peak power (cor-rectly decoded). Otherwise, the receiver detects a “0” bit for atime-spread pulse. The following sections elaborate our resultswith these InP-based O-CDMA components.

A. Mode-Locked Laser

An ultrafast light source based on semiconductor mode-locked (ML) laser is required as a transmitter in an O-CDMAintegrated photonic circuit. We have developed a 10-GHzcolliding-pulse mode-locked laser (CPM) [8], [9] in the platformfrom Fig. 1, allowing for the monolithic integration of the CPMlaser with other O-CDMA elements discussed in the remain-ing part of the paper. Compared to a regular mode-locked laserdesign, the CPM laser operates in the configuration where twosymmetrically counter-propagating pulses collide and bleachthe saturable absorber located at the center of the laser cavity,resulting in deeper saturation and more stable mode locking.

Fig. 5 shows the SEM image of a fabricated CPM laser.The required CPM cavity length for 10-GHz operation exceeds8 mm, which will require a fairly high drive current for all all-active design. Furthermore, long (>4 mm) all-active-mode MLlasers for achieving lower repetition rate tend to suffer fromstrong pulse shaping effects, resulting in poor chirping and jitterperformance [10]. Adopting the active-passive integration pro-cess in Fig. 1 allows for reducing the active section length tominimizing these effects and the drive current requirement [8].The active-passive interfaces are designed to be laterally tiltedat 45 ◦ relative to the waveguide direction. This lateral tilt iscritical for eliminating unwanted secondary pulses originatingfrom residual reflections at the interfaces, resulting in muchimproved mode-locking performance compared to the untilteddesigns. The active region has a total length of 2000 µm, di-viding into two gain sections and sandwiching the 45-µm-widesaturable absorber (SA) located at the center. The waveguideis further extended symmetrically on both sides with passivesections, forming the 8200-µm-long laser cavity.

The CPM laser can be synchronized to an external systemclock through either electrical hybrid mode locking (HML) [10],[11] at the fundamental or a subharmonic frequency, or throughthe optical synchronous mode locking (OSML) approach [12],[13]. In the electrical HML configuration, RF modulation signal


Fig. 6. (a) XFROG spectrogram and (b) extracted time-domain intensity andinsantaneous frequency of a nearly transform-limited pulse. (c) XFROG spec-trogram and (d) extracted time-domain intensity and instantaneous frequencyof a highly chirped pulse from the same CPM laser under different biasingconditions.

is applied to the saturable absorber of the CPM laser througha ground-signal-ground microwave probe, while the two gainsections and the SA are dc-biased through dc needle probes(forward current injection) and the microwave probe (reversebias voltage) through a bias-tee connection, respectively.

The CPM laser output pulse characteristics are very sensitivefunctions of the dc and RF biasing conditions. We utilized thecross-correlation frequency-resolved optical gating (XFROG)technique [14] (Fig. 9) for accurate extraction of the completetemporal amplitude and phase information, and the correspond-ing spectral domain profiles of the pulse simultaneously.

Fig. 6 shows the XFROG characterization of the CPM laserunder two different bias conditions. In Fig. 6(a), with 89-mAgain section and −7.5 V SA biasing, the tightly distributedXFROG spectrogram indicates that the pulse is close to beingtransform limited. The extracted time-domain intensity and in-stantaneous frequency profiles in Fig. 6(b) indicates a pulsewidth of 5.7 ps with minimal frequency chirp, with time-bandwidth product of 0.67 for a spectral width of 1.8 nm.Fig. 6(c) shows the XFROG trace of the same device at a differ-ent biasing condition with 156-mA gain section current, −4.5 Vacross the SA. The extracted time-domain profile in Fig. 6(d)shows that the resulting pulse is over 40-ps wide, with bothstrong linear and higher order chirp components. The XFROGtechnique allows the full characterization and systematic andefficient optimization of all the individual biasing parameters ofa CPM laser in real time.

We have previously reported in detail the electrical HML in-vestigation of the 10-GHz CPM laser [8]. By applying the RFclock signal matching the passive ML frequency of 10.3 GHzat 19 dBm, we obtained nearly transform-limited output withpulse width of 1.75 ps and time-bandwidth product 0.33, withrelatively low-timing jitter. This short-pulse low-chirp perfor-mance can be attributed to the careful balancing of the pulsebroadening and chirping effects inside the relatively short gain

Fig. 7. (a) RF power spectra at the fundamental ML frequency of the 10-GHzCPM laser under passive ML and optical injection locking conditions. Also,sampling scope traces of the CPM laser output triggered with the RF clocksource under (a) passive (b) optical injection locking conditions.

sections and the SA pulse truncating effects, respectively [15].Minimizing the active-passive interfacial reflections also playsan important role in achieving optimal pulse quality.

For synchronizing remote photonic-chip-based system nodesin an O-CDMA network, the electrical HML approach, whilerelatively easy to implement on a single device, requires sup-plying system clock signals at relatively high RF power lev-els at multiple distant locations, and is relatively cumbersometo implement. The alternative OSML approach would involvethe simple routing of a single optical clock signal through thenetwork fiber for injection locking of all CPM laser sourcessimultaneously.

For OSML optical injection locking characterization, a com-mercial tunable mode-locked fiber laser synchronized to an RFsynthesizer clock source provides the optical clock, which iscoupled into the 10-GHz CPM laser through a polarization-maintaining (PM) attenuator and a lensed fiber. The CPM out-put, collected through a lensed fiber from the other output facet,passes through a 3-nm tunable filter for removing any residu-als of the injection signal, and then routes to different instru-ments for time- and frequency-domain measurements, and thebit-error-rate tester (BERT) for BER measurements.

Fig. 7(a) shows the RF power spectra of a 10-GHz CPMlaser under both passive ML and injection locking conditions at−10 dBm coupled optical power, with the injection signal at1546 nm, 8 nm negatively detuned from the passive ML wave-length. Fig. 7(b) shows the corresponding 50-GHz optical sam-pling scope traces, triggered with the RF synthesizer clock.Compared to passive mode-locking, the RF spectral lineshapenarrows significantly in Fig. 7(a) under optical injection, corre-sponding to the sharply reduced timing jitter after synchroniza-tion. The estimated root mean square (rms) timing jitter of theinjection-locked CPM laser is 0.5 ps by integrating the single-side-band (SSB) phase noise spectrum, while the injected laserclock signal has an rms jitter of 0.17 ps. The effect of opticalinjection is primarily phase synchronization, without any ob-served changes in the pulse width and spectrum of the CPM laseroutput. Optical phase synchronization effects can be observedat injection power level as low as −23 dBm. The BER mea-surements also demonstrated that the CPM output is error-free


Fig. 8. (a) O-CDMA encoder chip layout, (b) encoder transmission spectrum,(c) packaged O-CDMA encoder chip, and (d) SEM picture of a laterally regrownpassive waveguide.

with BER <10−11 under 223 − 1 pseudorandom bit sequence(PRBS) data modulation, demonstrating that an optical injectionlocked 10-GHz CPM laser can indeed serve as a viable short-pulse light source in O-CDMA and other photonic integratedcircuit applications.

B. InP Encoder and Decoder

The previous integrated O-CDMA encoders are based oneither silica arrayed-waveguide gratings (AWG) with nonpro-grammable external phase shifters [16] or on ring microres-onators with relatively slow heater-based phase shifters [17].On the other hand, InP platform-based SPECTS O-CDMA(Fig. 1) encoders and decoders offer rapid code reconfigurationby employing electrooptical phase shifters and the possibility ofrealizing monolithically integrated O-CDMA transmitters andreceivers [5].

Fig. 8(a) shows the O-CDMA encoder mask layout. The AWGpair performs spectral demultiplexing and multiplexing. Phasemodulators [cross section in Fig. 2(b)] in between the AWGs ap-ply a phase shift, corresponding to the desired O-CDMA code,to each demultiplexed spectral channel. The input and outputwaveguides of the device are selected for optimal wavelengthmatch of the two AWGs. Fig. 8(b) shows the transmission spec-trum of an O-CDMA encoder with 32-dB fiber-to-fiber insertionloss. The loss can still be greatly reduced by improving the fibercoupling (here, 8-dB per fiber-to-chip coupling) and the fabri-cation process. Fig. 8(c) and (d) show the packaged chip and ascanning electron micrograph (SEM) picture of the cross sectionof the chip.

One of the challenges is to calibrate the phase response of theindividual channels of the InP chip for two reasons. First, thephase-shifter response must be known as a function of injectioncurrent or reverse-bias voltage (both can be used to generate aphase shift) to apply the right amount of phase shift. Althoughphase shifter performance can be extracted using test structures,data from the actual shifters in the encoder is preferable due to

Fig. 9. Setup for XFROG-based phase-error measurement.

Fig. 10. (a) Phase and (b) intensity response of phase-compensated encoder.

possible variations between shifters. Second, the chip containsphase errors in the eight channels due to fabrication tolerances,causing a phase offset. For channel spacings up to ∼40 GHz(note the encoder chip has 180 GHz), the differential phase ofthe channels can be extracted by detecting the coherent beatingbetween two spectral channels of the encoder [18]. This ap-proach requires the precise alignment of two axial modes froma mode-locked laser signal with two encoder channels. The twomodes produce a coherent beating response at the encoder out-put. However, this beating is hard to detect beyond 40 GHzwithout sophisticated high-speed electronics. In contrast, wecharacterized the phase response of the 180-GHz encoder chipusing the X-FROG technique [19], which avoids the preci-sion spectral alignment requirements in the coherent beatingapproach. Also, the phase information from all channels is ac-quired in a single measurement.

Fig. 9 shows the principle of X-FROG-based phase-errormeasurement. The ML laser and compressor generated 0.5-pspulses with a repetition rate of 10 GHz. A polarization con-troller (PC) put the pulses in TE polarization before a lensedfiber injected them into the encoder chip at 10 dBm. At the chipoutput, an erbium-doped fiber amplifier (EDFA) amplified theoutput pulse and fed it into the X-FROG through a PC. Theamplified uncompressed pulse of the mode-locked laser servedas reference for the X-FROG.

Fig. 10(a) shows the retrieved phase information (curve at thebottom corresponds to power in the channels), and Fig. 10(b)


Fig. 11. Experimental setup for 6 × 10 Gb/s O-CDMA testbed.

shows the encoder output. The dotted lines in Fig. 10(a) and (b)are without phase-error compensation, solid lines with (byreverse-biasing the phase shifters). After compensation, thephases of all channels (peak of solid lines) are almost flat[Fig. 10(a)]. The compensation clearly improves the peak powerand signal-to-noise (SNR) ratio [Fig. 10(b)]. The ringing peaksat 5.5-ps spacing originate from the 180-GHz comb-shapedspectral response of the AWG [Fig. 8(b)]. This phase com-pensation result shows both the proper operation of the InPencoder and the value of FROG as encoder characterizationtool.

Fig. 11 shows the schematic of the O-CDMA testbed. Themode-locked fiber laser source generated a 500-fs pulse train ata 10-GHz repetition rate. A LiNbO3 modulator modulated thepulse train with a 10 Gb/s 231 − 1 PRBS pattern. A time-muxsplitted the modulated train into two time slots of 50 ps each.A dispersion-compensated erbium-doped fiber amplifier (DC-EDFA) boosted the train, which is subsequently spatially splitinto three branches. One branch passes through the InP encoder[20] as the intended user (user channel). The other two branchesare encoded by spatial light modulator (SLM) encoders [6],which acted as independent interferers (IF1 and IF2). The SLMencoders contained variable time delays for synchronization.After encoding, the intended user and interferers were combinedand routed to the InP decoder. Upstream to each chip, an EDFAfollowed by a polarization controller supplied 21 dBm of TEpolarized light.

The InP devices each consist of an identical pair of 8-channelAWGs with 180-GHz channel spacing and a free-spectral range(FSR) of 12-channel spacings. The 1.4-THz wavelength spanprovided by the eight channels is sufficient for encoding sub-picosecond (∼0.5 ps) pulses. The InP devices are packaged ina standard 14-pin butterfly package with temperature control,which allows for an exact wavelength match of the encoderand decoder. The 2000-µm-long phase-shifters are controlledby electrically biasing wire-bonded electrical leads.

We selected the 8-chip Walsh code [21] set for our spectralphase-coding experiment. The InP encoder applies code W1containing the bits[11111111], whereas the InP decoder appliesconjugate code (*) W1* [00000000]. Here, 1 denotes π radiansphase shift and 0 denotes zero phase shift in the respective phase

Fig. 12. Time-domain cross-correlation traces of the correctly decoded signalfor (a) without coding, (b) 2 users, (c) 4 users and (d) 6 users.

shifters. SLM interferers IF1 and IF2 apply W2 [10101010] andW4 [10011001], respectively. These codes were selected for op-timal suppression of the multiuser interference (MUI). The gate[a nonlinear optical loop mirror (NOLM)], to be replaced by theintegrated MZI, is followed by power thresholding with highlynonlinear fiber (HNLF) and a long-pass filter. This combina-tion suppresses an incorrectly decoded signal while it passes acorrectly decoded signal.

Fig. 12 displays the effect of MUI on the pulse output.Fig. 12(a) shows the cross-correlation trace after the encoder–decoder pair with phase-error compensation but without coding.The∼3-ps time gating around t = 0 ps (dotted line) removes theringing peaks at the expense of signal power. Fig. 12(b) showsthe cross-correlation trace of the output of the encoder–decoderpair without interferers (extra users), with W1 and W1* codesfor encoder and decoder for two time slots. Fig. 12(c) and (d)show the traces with added interferer W2 and interferers W2and W4, respectively (two time slots).

Fig. 13(a) shows the BER measurement results for the 6-userO-CDMA system with the InP encoder and decoder. Back-to-back data were taken with the O-CDMA encoder and decoderpair bypassed. The received power in all BER curves is definedas the average input power per user as measured at the inputof the “O-CDMA receiver.” The 2-user and 4-user experimentsdemonstrated error-free operation and a power penalty of 5.5 dBat BER = 10−9 with respect to back-to-back data. The penaltycan be mainly attributed to the power lost to the ringing peaks.The 6-user case attained an error floor at 10−9, primarily dueto MUI inside the time gate. The MUI effects are relativelysmall for 2 and 4 users, resulting in nonobservable noise floor.For a narrower time gate, window and/or longer code lengthwould mitigate the MUI effects significantly and an error-freeoperation is expected.

In order to investigate the effect of the spectral filter-ing and loss in the InP-based en/decoders, we repeated theBER measurement but replaced the InP devices with SLM-based en/decoders. The SLM devices approach an ideal coding


Fig. 13. BER measurement results for (a) InP encoder and decoder chips and(b) SLM encoders and decoder.

performance due to their rectangular filter shape. Fig. 13(b)shows the BER measurement result and eye diagrams. The2-user and 4-user experiments demonstrated error-free oper-ation with a power penalty of 2 and 5 dB, respectively, atBER = 10−9. The error floor below 10−9 in the 6-user SLM caseconfirms that MUI, and not the InP chip response, is the domi-nant limitation for error-free operation of the O-CDMA system.

C. MZI Threshold Detector Development

The last step of the SPECT O-CDMA transmission link re-quires nonlinear optical thresholding: the detection of properlydecoded optical pulses with narrow pulse width and high peakpower, and the blocking of the time spread by improperly de-coded interferer signal. Compared to other nonlinear-fiber-basedthresholder approaches, the MZI thresholder is less bulky andmore sensitive [22], [23] and is compatible with monolithicintegration with other semiconductor components. Here, we de-scribe an improved “clock-gated” MZI scheme with respect tothe “self-gated” approach reported previously [5]. Both schemesdepend on differential operation of the MZI.

Fig. 14 shows the self-gated and clock-gated MZI operation.In the self-gated scheme, no external clock signal is provided.The decoded pulse is split into two, and each part enters adistinct MZI arm with a time delay. Only the correctly decodedpulses have sufficiently fast rising edges to saturate the nonlinearoptical elements in the MZI arms, opening a differential gate for

Fig. 14. O-CDMA thresholder in (a) nongated and (b) clock-gated configura-tion.

Fig. 15. Graph of power ratio after MZI thresholder for self-gated and clock-gated operations.

the CW probe signal. The clock-gated scheme uses the externalshort-pulse clock signal for generating the differential gate ofthe MZI.

Fig. 15 shows the simulated discrimination ratio (the ratioof the integrated energy of a correctly decoded pulse versusthat of an incorrectly decoded pulse) as a function of the timedelay between the two MZI arms for self-gated and clock-gatedoperation using 64-bit Walsh codes. Clock-gating outperformsself-gating by up to 20 dB, because in the self-gated condition,a partial broad gate can still be present for incorrectly decodedsignal, resulting in incomplete blocking.

For improved performance, we investigated a new class ofintegrated MZI based on reverse-biased EA [24]. The reverse-biased EA sections result in faster carrier sweepout (tens of pi-coseconds) and less amplified stimulated emission (ASE) noisecompared to the conventional forward-biased SOAs in the MZIarms favoring high speed and lower power consumption opera-tion with greater cascadability.

Fig. 16(a) shows the fabricated EA-MZI with four multi-mode interference (MMI)-based 3-dB couplers, two active EAsections of 200 and 50 µm in length, and two 600-µm-longphase shifters (see also Fig. 2). The phase modulators allowthe MZI to be biased for optimal destructive interference atthe MZI output, whereas the unequal EA section lengths allowbetter power balance between the two arms (without pump).Injection of an optical pump signal into the shortest EA inducescross phase and amplitude modulation, and the device operatesas a “noninverting” optical switch (or wavelength converter).

Fig. 16(b) shows the normalized MZI switching output ver-sus pump input power with the pump signal at 1540 nm, probesignal at 1550 nm, and the upper and lower EA reverse biased at−1 and −12 V, respectively. A 10-dB pump power change (ER)resulted in 18-dB output switching ratio, demonstrating the op-tical signal regeneration capability of the EA-MZI. Fig. 16(b)


Fig. 16. (a) Photograph of fabricated EA-MZI with active-passive integratedwaveguide. b) Normalized MZI transmission versus pump input power, insetshowing EA-MZI output at 2.5 Gb/s.

inset shows the MZI wavelength conversion output from a2.5 Gb/s 27 − 1 PRBS pump input. These results demonstratethe optical switching capacity of the EA-MZI device, and its po-tential as a high-speed, low-power, and low-noise optical thresh-older element.

IV. FUTURE WORK

It is a challenging task to achieve a monolithically integratedon-chip system. To make the on-chip O-CDMA system operateproperly, several key issues need to be addressed. These includelow-loss waveguide, the realization of narrow channel spacingAWG, on-chip mirror for lasing operation of the mode-lockedlaser. By optimizing the layer stack of the material system of thedevice and processing procedure of the fabrication, we achieved3 dB/cm for passive waveguide for doped material. This low-loss waveguide combined with on-chip amplification by usingSOA can satisfy the requirement for integrated system. We willdiscuss two other issues in this section.

A. Realization of Narrow Channel Count AWG

For O-CDMA, the amount of users can be increased if the(AWG) channel spacing is reduced or if the bandwidth of thelaser pulses is increased. The narrower the channel spacing,the longer the code length that can be applied to allow bettersuppression of MUI [5]. Currently, we are working toward 50and 20 GHz channel spacing encoders. As an intermediate step,we have developed a 100-GHz channel spacing AWG to opti-mize the layer stack, fabrication the processing step and AWGdesign parameters (such as input/output free-propagation regionand waveguide shape).

Fig. 17. Transmission spectra of all ten transmission channels of 100-GHzchannel spacing AWG.

Fig. 18. (a) Optical output of etched facet laser versus injection current (1%duty cycle) and (b) FIB-etched photonic crystal laser mirror.

Fig. 17 shows the transmission spectrum of a 10 channel ×100 GHz channel spacing AWG. The FSR of this AWG is12.8 nm, which is the same as our 8-channel device. The3-dB bandwidth of single channel is 0.35 nm. The fiber-to-fiberloss of this single AWG is 16 dB. The SNR is more than 25 dB,and side-lobe suppression is about 20 dB. The channel spacingis 0.8 nm. The central wavelength (channel 6) is 1547.7 nm.The nonuniformity among 10 channels is less than 2 dB. Com-pared to 8-channel× 200 GHz channel spacing device, the fiber-to-fiber loss of the 100-GHz design does not increase. Thisproves that the improved design allows narrower channel spac-ings without performance loss, making higher channel countAWGs feasible. Furthermore, we have characterized the phaseerrors in the AWG arms using a new method: XFROG in com-bination with an inverse Fourier transform. Knowledge of theseerrors will enable us to reduce systematic phase errors in theAWG arms.

B. On-Chip Mirror for Mode- Locked Laser Integration

Monolithic integration of the CPM laser source with otherelements requires the development of built-in mirror reflectors[Fig. 2(d)]. We are developing the photonic crystal (PC)-basedlaser mirrors similar to that demonstrated on CW semiconduc-tor lasers [25]. The etched mirrors allow for accurate control ofthe repetition rate. Fig. 18(b) shows the photonic crystal mir-rors fabricated using focused-ion-beam (FIB) and titanium etchmask. Fig. 18(a) displays pulsed laser output for three laser cavi-ties having different lengths. Lasing threshold currents observedwere around 25 mA (1% duty cycle). The analyses of the Fourier


components of the output spectra confirmed lasing action in thecavity defined by the etched mirrors instead of another possiblecavity defined between the antireflection-coated cleaved facets.

V. CONCLUSION

InP platforms have strong potentials to provide compact andstable monolithic integration for system scale applications. Wepresented two InP integration platforms and demonstrated upto 60 Gb/s O-CDMA using InP encoders/decoders. Adding ourresults on the 10 Gb/s mode-locked laser transmitter with nearlytransform-limited 1.75-ps pulses and the EA-based MZI thresh-old detector, we firmly believe in the feasibility of a singleO-CDMA transceiver chip with transmission capacity beyond100 Gb/s. Future work involves improved channel count bydenser spectral resolution of the O-CDMA system and shortermode-locked laser pulses for wider pulse spectra.

REFERENCES

[1] E. J. Skogen, J. W. Raring, G. B. Morrison, C. S. Wang, V. Lal, M.L. Masanovic, and L. A. Coldren, “Monolithically integrated active com-ponents: A quantum-well intermixing approach,” IEEE J. Sel. Top. Quan-tum Electron., vol. 11, no. 2, pp. 343–355, Mar./Apr. 2005.

[2] J. J. M. Binsma, J. H. den Besten, and R. G. Broeke, “InP-based pho-tonic integration technology,” presented at the Opt. Amplifiers TheirAppl./Integr. Photon. Res. Topical Meet., 2004, Paper IFB1.

[3] E. Bente and M. Smit, “Ultra fast InP integrated circuits,” Proc. SPIE,vol. 6124, pp. 612419–612444, 2006.

[4] S. R. Forrest, M. R. Gokhale, P. V. Studenkov, and F. Xia, “Integrated pho-tonics using asymmetric twin-waveguide structures,” in Proc. Int. Conf.Indium Phosphide Related Mater., 2000, pp. 13–16.

[5] C. Ji, R. G. Broeke, Y. Du, J. Cao, N. Chubun, P. Bjeletich, F. Olsson,S. Lourdudoss, R. Welty, C. Reinhardt, P. L. Stephan, and S. J. B. Yoo,“Monolithically integrated InP based photonic chip development for O-CDMA systems,” IEEE J. Sel. Top. Quantum Electron., vol. 11, no. 1,pp. 66–77, Jan./Feb. 2005.

[6] V. J. Hernandez, W. Cong, J. Hu, C. Yang, N. K. Fontaine, R. P. Scott,Z. Ding, B. H. Kolner, J. P. Heritage, and S. J. B. Yoo, “A 320-Gb/scapacity (32-User × 10 Gb/s) SPECTS O-CDMA network testbed withenhanced spectral efficiency through forward error correction,” J. Lightw.Technol., vol. 25, no. 1, pp. 79–86, Jan. 2007.

[7] R. P. Scott, N. K. Fontaine, J. P. Heritage, B. H. Kolner, and S. J.B. Yoo, “3.5-THz wide, 175 mode optical comb source,” presented atthe IEEE/OSA Opt. Fiber Commun. Conf., 2007, Paper OWJ3.

[8] C. Ji, N. Chubun, R. G. Broeke, J. Cao, Y. Du, S. J. B. Yoo, K. Y. Liou, J.R. Lothian, S. Vatanapradit, S. N. G. Chu, B. Patel, W. S. Hobson, and W.T. Tsang, “Synchronized transform-limited operation of 10-GHz collidingpulse mode-locked laser,” IEEE Photon. Technol. Lett., vol. 18, no. 4,pp. 625–627, Feb. 2006.

[9] Y. K. Chen and M. C. Wu, “Monolithic colliding-pulse mode-lockedquantum-well lasers,” IEEE J. Quantum Electron., vol. 28, no. 10,pp. 2176–2185, Oct. 1992.

[10] K. Yvind, D. Larsson, L. J. Christiansen, J. Mork, J. M. Hvam, andJ. Hanberg, “High-performance 10 GHz all-active monolithic modelockedsemiconductor lasers,” Electron. Lett., vol. 40, no. 12, pp. 735–737, Jun.2004.

[11] H. Fan, C. Wu, M. El-Aasser, N. K. Dutta, U. Koren, and A. B. Piccirilli,“Colliding pulse mode-locked laser,” IEEE Photon. Technol. Lett., vol. 12,no. 8, pp. 972–973, Aug. 2000.

[12] S. Arahira and Y. Ogawa, “480-GHz subharmonic synchronous modelocking in a short-cavity colliding-pulse mode-locked laser diode,” IEEEPhoton. Technol. Lett., vol. 14, no. 4, pp. 537–539, Apr. 2002.

[13] H. Kurita, T. Shimizu, and H. Yokoyama, “Experimental investigationsof harmonic synchronization conditions and mechanisms of mode lockedlaser diodes by optical pulse injection,” IEEE J. Sel. Top. QuantumElectron., vol. 2, no. 3, pp. 508–513, Sep. 1996.

[14] R. Trebino, Frequency Resolved Optical Gating: The Measurement OfUltrashort Laser Pulses. Boston, MA: Kluwer, 2000.

[15] S. Gee, R. Coffie, P. J. Delfyett, G. Alphonse, and J. Connolly, “Intracavitygain and absorption dynamics of hybrid modelocked semiconductor lasersusing multiple quantum well saturable absorbers,” Appl. Phys. Lett.,vol. 71, pp. 2569–2571, 1997.

[16] H. Tsuda, H. Takenouchi, T. Ishii, K. Okamoto, T. Goh, K. Sato, A. Hirano,T. Kurokawa, and C. Amano, “Spectral encoding and decoding of 10 Gbit/sfemtosecond pulses using high resolution arrayed-waveguide grating,”Electron. Lett., vol. 35, no. 14, pp. 1186–1188, Jul. 1999.

[17] A. Agarwal, P. Toliver, R. Menendez, S. Etemad, J. Jackel, J. Young,T. Banwell, B. E. Little, S. T. Chu, W. Chen, W. Chen, J. Hryniewicz,F. Johnson, D. Gill, O. King, R. Davidson, K. Donovan, and P. J. Delfyett,“Fully programmable ring-resonator-based integrated photonic circuit forphase coherent applications,” IEEE J. Lightw. Technol., vol. 24, no. 1,pp. 77–87, Jan. 2006.

[18] A. Agarwal, P. Toliver, R. Menendez, S. Etemad, J. Jackel, J. Young,T. Banwell, B. E. Little, S. T. Chu, and P. Delfyett, “Fully-programmablering resonator based integrated photonic circuit for phase coherent appli-cations,” presented at OFC, 2005, Paper PDP6.

[19] R. G. Broeke, J. Cao, N. K. Fontaine, C. Ji, N. Chubun, F. Olsson,S. Lourdudoss, B. H. Kolner, J. P. Heritage, and S. J. B. Yoo, “Phasecharacterization of an inp based optical-cdma encoder using frequency-resolved optical gating (FROG),” presented at the IEEE LEOS Annu.Meet., 2005, Paper ML2.

[20] J. Cao, R. G. Broeke, N. Fontaine, C. Ji, Y. Du, N. Chubun, K. Aihara,A.-V. Pham, F. Olsson, S. Lourdudoss, and S. J. B. Yoo, “Demonstrationof spectral phase O-CDMA encoding and decoding in monolithically inte-grated arrayed-waveguide-grating-based encoder,” IEEE Photon. Technol.Lett., 2007, to be published.

[21] F. J. MacWilliams and N. J. A. Sloane, The Theory of Error CorrectingCodes. New York: North-Holland, 1977.

[22] H. P. Sardesai and A. M. Weiner, “Nonlinear fibre-optic receiver for ul-trashort pulse code division multiple access communications,” Electron.Lett., vol. 33, no. 7, pp. 610–611, Mar. 1997.

[23] Z. Jiang et al., “Spectrally coded O-CDMA system with four users at2.5 Gbit/s using low power nonlinear processing,” Electron. Lett., vol. 40,no. 10, pp. 623–625, May 2004.

[24] M. Suzuki, H. Tanaka, and S. Akiba, “Effect of hole pileup at heteroint-erface of modulation voltage in GaInAsP electroabsorption modulators,”Electron. Lett., vol. 25, no. 2, pp. 88–89, Jan. 1989.

[25] J. Moosburger, M. Kamp, F. Klopf, J. P. Reithmaier, and A. Forchel,“Semiconductor lasers with 2D photonic crystal mirrors based on a wet-oxided Al2O3-mask,” IEEE Photon. Technol. Lett., vol. 13, no. 5, pp. 406–408, May 2001.

Ronald G. Broeke received the M.S. degree in physics from the Universityof Utrecht, Utrecht, The Netherlands, in 1998, and the Ph.D. degree from theDepartment of Electrical Engineering, Delft University of Technology, Delft,The Netherlands in 2003.

He is currently a Postdoctoral Research Scientist with the Department ofElectrical and Computer Engineering, University of California, Davis.

Jing Cao (S’01) received the B.S. and M.S degrees from the Department ofElectronics Engineering, Tsinghua University, Beijing, China in 1997 and 2000,respectively. He is currently working toward the Graduate degree at the De-partment of Electrical and Computer Engineering, University of California,Davis.

His current research interests include optical integrated devices and systemintegration for next generation optical network.

Mr. Cao is a member of the Optical Society of America.

Chen Ji received the B.S. degree in physics from the University of Illinois,Urbana-Campaign, Urbana, in 1993, and the Ph.D. degree in electrical engi-neering from Cornell University, Ithaca, NY, in 1999.

He is currently a Postdoctoral Research Scientist with the Department ofElectrical and Computer Engineering, University of California, Davis. His re-search interests include semiconductor lasers, optical integrates devices, andsystem integration for next generation optical networks.


Sang-Woo Seo (M’03) received the B.S. degree inelectrical and computer engineering from Ajou Uni-versity, Suwon, Korea, in 1997, the M.S degree ininformation and communications from Kwangju In-stitute of Science and Technology (KJIST), Kwangju,Korea, in 1999, and the Ph.D. degree in electrical andcomputer engineering from Georgia Institute of Tech-nology, Atlanta, in 2003.

From 2003 to 2005, he was a Research Engineerwith Georgia Institute of Technology and Duke Uni-versity, Durham, NC. Since 2005, he has been with

the University of California, Davis. His current interests include high-speedmodulators, photonic integrated devices, and system integration for optical codedivision access network and optical arbitrary waveform generation.

Yixue Du (S’03) received the M.S. degree in physics from Oklahoma StateUniversity, Stillwater, OK, in 1998. She is currently working toward the Ph.D.degree in electrical engineering at the University of California, Davis.

Her research interests include simulation of the network behavior, and thedesign and testing of nonlinear optical thresholding device in optical codedivision multiple access networks.

Nick K. Fontaine received the B.S. degree in electrical engineering and opticalengineering from the University of California, Davis, in 2004, where he iscurrently working toward the Ph.D. degree.

His research interests include frequency resolved optical gating, opticalarbitrary waveform generation, optical code-division multiple access, and verysmall lasers.

Jong-Hwa Baek received the M.S. and Ph.D. degreesin physics from Korea Advanced Institute of Scienceand Technology, Daejeon, Korea, in 1999 and 2004,respectively.

He is currently a Research Scientist with the Uni-versity of California, Davis. His research interestsinclude photonic integrated circuits, photonic crystalactive devices, and nanophotonic technologies.

John Yan received the B.S. degree in electrical engi-neering from the University of California, Davis, in2005, where he is currently working toward the Ph.D.degree.

His research interests include optical semiconduc-tor integration, radio-frequency ICs and mixed-signalICs.

Mr. Yan is a member of the Tau Beta Pi.

Francisco M. Soares received the M.S. degree in electrical engineering from theDelft University of Technology, The Netherlands, in 2000. In 2006, he receivedthe Ph.D. degree from the Department of Electrical Engineering, TechnicalUniversity of Eindhoven, Eindhoven, The Netherlands.

Since 2006, he is with the University of California, Davis, where he is engagedin research on optical arbitrary waveform generators in InP technology. Hisresearch interests include theories, modeling, characterization, and fabricationof compound semiconductor devices such as optical switches and modulators,arrayed waveguide gratings, spot-size converters, and multimode interferencecouplers.

Fredrik Olsson received the B.S. degree in physics from the Royal Institute ofTechnology (KTH), Stockholm, Sweden, in 2000, where he is currently workingtoward the Ph.D. degree at the School of Information and CommunicationTechnology.

His research interests include integration of III/V semiconductor materialson silicon and growth of optical integrated devices.

S. Lourdudoss (M’97–SM’99) received the M.S. degree in chemistry fromMadras University, Chennai, India, in 1976 and the Ph.D. degree in chemistryfrom Faculte Libre des Sciences de Lille, Lille, France, in 1979.

He was a Postdoctoral Fellow at the Royal Institute of Technology (KTH),Stockholm, Sweden, where he worked on the thermochemical aspects of ther-mal energy storage and absorption heat pumps until 1985. He then joined theSwedish Institute of Microelectronics, Stockholm. Currently, he is a Profes-sor of semiconductor materials at KTH, and is involved in undergraduate andgraduate education. His research interests include MOVPE of GaN and relatedmaterials, and the issues related to the monolithic integration of III–V materialson a single platform as well as on Si by HVPE.

Anh-Vu H. Pham (SM’03) received the B.E.E. (withhonors), M.S., and Ph.D. degrees from the GeorgiaInstitute of Technology, Atlanta, in 1995, 1997, and1999, respectively.

He cofounded RF Solutions, LLC, Warren, NJ, in1997, which was acquired by Anadigics in 2003. Hehas held Faculty position with Clemson University,Clemson, SC. He is currently an Associate Profes-sor with the University of California, Davis. He hascoauthored about 80 technical papers published inseveral journals and conference proceedings. His re-

search interests include RF and high-speed packaging and signal integrity, andRFIC design.

Prof. Pham serves as a member of the IEEE Microwave Theory and Tech-niques Society (IEEE MTT-S) International Microwave Symposium’s TechnicalProgram Committee on Power Amplifiers and Integrated Circuits. He has beenthe Chair of the IEEE MTT-12 Microwave and Millimeter Wave Packaging andManufacturing Technical Committee of the IEEE MTT-S. He was the recipientof the 2001 National Science Foundation CAREER Award.

Michael Shearn received the B.S. degrees in elec-trical engineering, physics, and mathematics fromSouthern Methodist University, Dallas, TX, in 2004.He is currently working toward the Ph.D. degree inapplied physics at the California Institute of Technol-ogy, Pasadena.

His research interests include silicon photonics,optical resonators, and novel fabrication techniques.

Axel Scherer received the B.S., M.S., and Ph.D. de-grees from the New Mexico Institute of Mining andTechnology, Socorro, New Mexico, in 1981, 1982,and 1985, respectively.

From 1985 to 1993, he was with the Quantum De-vice Fabrication Group, Bellcore, Morristown, NJ.He is currently the Bernard E. Neches Professor ofElectrical Engineering, Applied Physics, and Physicswith the California Institute of Technology, Pasadena,specializing in device microfabrication. His researchinterests include design and fabrication of functional

photonic, nanomagnetic, and microfluidic devices.


S. J. Ben Yoo (S’82–M’84–SM’97) received theB.S., M.S., and Ph.D. degrees in electrical engineer-ing from Stanford University, Stanford, CA, in 1984,1986, and 1991, respectively.

He is currently a Professor of electrical engi-neering with the University of California, Davis(UC Davis), where he is also the Director of Cen-ter for Information Technology Research in theInterest of Society. His research interests includehigh-performance all-optical devices, systems, andnetworking technologies for the next generation In-

ternet; and architectures, systems integration, and network experiments relatedto all-optical label switching routers and optical code division multiple accesstechnologies. Prior to joining UC Davis in 1999, he was a Senior ResearchScientist with Bell Communications Research, Morristown, NJ, leading tech-nical efforts in optical networking research and systems integration. He was

with Stanford University, Stanford, CA, before joining Bell CommunicationsResearch. He also conducted research on lifetime measurements of intersub-band transitions and on nonlinear optical storage mechanisms at Bell Lab-oratories, Murray Hill, NJ, and IBM Research Laboratories, San Jose, CA,respectively.

Prof. Yoo is a Fellow of the IEEE Lasers and Electro-Optics Society and aFellow of the Optical Society of America. He is a member of the Tau Beta Pi.He was the recipient of the DARPA Award for Sustained Excellence in 1997,the Bellcore CEO Award in 1998, and the Mid-Career Research Faculty Award(UC Davis) in 2004. He was the Co-Chair of the Technical Program Commit-tee for APOC 2004 and APOC 2003, and was a member of technical programcommittees for several conferences. He is the General Co-Chair for IEEE LEOSPhotonics in Switching Conference 2007. He is an Associate Editor for IEEEPHOTONICS TECHNOLOGY LETTERS, and a Guest Editor for IEEE/OSA JOUR-NAL OF LIGHTWAVE TECHNOLOGY in 2005 and IEEE JOURNAL OF SELECTED

TOPICS IN QUANTUM ELECTRONICS in 2007.

IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM...

Documents

Transcript of IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM...